Sodium Alginate Toughening of Gelatin Hydrogels

Samples were characterized using ramp-hold stress relaxation tests. The experimental data from these tests was then fit to the Generalized Maxwell (GM...
23 downloads 3 Views 2MB Size
Download PDF
Article Cite This: ACS Biomater. Sci. Eng. 2017, 3, 3176−3182

Sodium Alginate Toughening of Gelatin Hydrogels Michael A. Samp, Nicolae C. Iovanac, and Adam J. Nolte* Department of Chemical Engineering, Rose-Hulman Institute of Technology, 5500 Wabash Avenue, Terre Haute, Indiana 47803, United States S Supporting Information *

ABSTRACT: Gelatin is a popular material for the creation of tissue phantoms due to its ease-of-use, safety, low relative cost, and its amenability to tuning physical properties through the use of additives. One difficulty that arises when using gelatin, especially in low concentrations, is the brittleness of the material. In this paper, we show that small additions of another common biological polymer, sodium alginate, significantly increase the toughness of gelatin without changing the Young’s modulus or other low-strain stress relaxation properties of the material. Samples were characterized using ramp-hold stress relaxation tests. The experimental data from these tests were then fit to the Generalized Maxwell (GM) model, as well as two models based on a fractional calculus approach: the Kelvin−Voigt Fractional Derivative (KVFD) and Fractional Maxwell (FM) models. We found that for our samples, the fractional models provided better fits with fewer parameters, and at strains within the linear elastic region, the linear viscoelastic parameters of the alginate/gelatin and pure gelatin samples were essentially indistinguishable. When the same ramp-hold stress relaxation experiments were run at high strains outside of the linear elastic region, we observed a shift in stress relaxation to shorter time scales with increasing sodium alginate addition, which may be associated with an increase in fluidity within the gelatin matrix. This leads us to believe that sodium alginate acts to enhance the viscosity within the fluidic region of the gelatin matrix, providing additional energy dissipation without raising the modulus of the material. These results are applicable to anyone desiring independent control of the Young’s modulus and toughness in preparing tissue phantoms, and suggest that sodium alginate should be added to low-modulus gelatin for use in biological and medical testing applications. KEYWORDS: gelatin, alginate, toughness enhancement, force relaxation, viscoelastic modeling, fractional calculus

■

INTRODUCTION Owing to the difficulty or impracticality of using biological tissue samples in medical procedure testing and training, in many instances the tissue is substituted with a phantom whose specific physical properties may be tuned to match those of the biological sample.1 The decision on which material to use for a phantom is highly dependent on the type of testing to be performed, as the nature of the tests dictates which material properties are important. For applications involving the physical deformation of the sample, ranging from ultrasonic imaging to tissue palpations, the viscoelastic properties of the material are of particular importance. While many soft materials will exhibit some degree of viscoelastic behavior, biological tissues are often complex heterogeneous structures, and as such may display viscoelastic characteristics that cannot easily be reproduced in many testing materials.1 As such, a number of material candidates have been proposed, each with its own advantages and disadvantages.2 Despite their relative simplicity, gelatin-based phantoms have long provided a good starting material for basic tissue modeling.3 The primary advantages of using gelatin over other casting materials lie in its low cost, ease of use,4 and safety.5 Additionally, the use of gelatin allows for easy tuning of material parameters, © 2017 American Chemical Society

such as the Young’s modulus and/or density of the sample, by simply varying the concentration of gelatin in solution,6 or by including additives.3 Despite their advantages, one significant drawback found in gelatin-based phantoms is their frangibility, particularly in low modulus samples.7 The toughness of the samples can be increased by raising the concentration of gelatin, but this comes at the cost of increasing the Young’s modulus as well.6 As others have previously noted, in order to accurately mimic low-modulus tissues, it is often necessary to develop procedures that allow one to independently control certain mechanical properties within the phantom material.8 Others have shown that blending sodium alginate into gelatin can provide an increase in toughness for a dried film.9 In this paper we demonstrate that the addition of minority amounts of sodium alginate to aqueous-based low-modulus gelatin hydrogels can lead to an increase in the toughness of the material without appreciably affecting the Young’s modulus, thus suggesting these materials may serve as facile, low-modulus tissue phantom systems. Received: May 23, 2017 Accepted: November 12, 2017 Published: November 13, 2017 3176

DOI: 10.1021/acsbiomaterials.7b00321 ACS Biomater. Sci. Eng. 2017, 3, 3176−3182

Article

ACS Biomaterials Science & Engineering

Figure 1. (a) n-Armed Generalized Maxwell (GM) model with n viscosities and n + 1 elastic moduli. (b) Three-term fractional Kelvin Voigt (KVFD) model. (c) Three-term Fractional Maxwell (FM) model. The diamonds represent fractional-order viscoelastic elements characterized by parameters η and α.

Figure 2. Plots of (a) toughness, (b) elastic modulus, and (c) strain-to-break as a function of alginate concentration. The height of each bar represents the average of all trials and the uncertainty bars represent the standard error in the average measurement. For these results, seven samples were tested at each alginate concentration. These tests were performed with a strain rate ramp of 0.5% s−1.

elastic moduli and viscosities, corresponding to the numbers of springs and dashpots, respectively, to model stress relaxation. We additionally considered the Kelvin−Voigt Fractional Derivative (KVFD) and Fractional Maxwell (FM) models, which consist of a single spring connected either in parallel or in series, respectively, with a fractional-order element displaying properties of an intermediate character between pure elastic and viscous behavior as shown in Figure 1 panels b and c. Such elements utilize a mathematical operator known as a fractional derivative, where the order of the derivative, α, is permitted to take on real number values between 0 and 1. The stress on the fractional-order element (σf) is related to the strain (εf) by eq 3.11

Linear viscoelastic modeling is a tool that can shed light on the mechanisms that give rise to the particular mechanical characteristics of the material. These models typically make use of several fitting parameters that are conceptually associated with arrangements of springs and dashpots, which reflect respectively the elastic (through Young’s moduli, Ei) and viscous (through viscosities, ηi) response mechanisms within the material.10 These responses are captured through the use of constitutive relationships that relate the stress (σi) and strain (ϵi) on each model element as follows:10 σi = Ei ϵi (for an elastic, or “Hookean” material response) (1)

σi = ηi

dϵi (for a viscous, or “Newtonian” material response) dt

σf = η

(2)

dα ϵ f dt α

(3)

By introducing a time constant for the fractional-order element, τ, as η = Eτα, eq 3 can be rewritten as

Three modeling schemes were considered for this study. The first was the Generalized Maxwell (GM) model, shown in Figure 1a, which comprises series and parallel arrangements of spring and dashpot elements. Depending on the number of elements provided to the model, the GM model uses varying numbers of

σf = Eτ α 3177

dα ϵ f dt α

(4) DOI: 10.1021/acsbiomaterials.7b00321 ACS Biomater. Sci. Eng. 2017, 3, 3176−3182

Article

ACS Biomaterials Science & Engineering

until a reading of 0 N was achieved. After this precycle, the sample was allowed to recover for 2 min before the start of the actual experiment. To determine the toughness, samples were compressed at a strain rate of 0.5% s−1 until failure was observed. The strain rate for toughness testing was chosen to coincide with the rate used for relaxation tests, as described below. Since these tests were performed using parallel plate compression, after the material fractured, it did not slide out from under the testing head and continued to be compressed. Because of this, the maximum stress on the stress−strain curve did not always correspond to the fracture strength of the material. Instead, the failure strain was determined manually as the strain where the stress suddenly deviated downward from its smooth upward trend, which often coincided with the visual observation of cracks forming on the sample surface. Because it was difficult to guarantee that we could always visualize the onset of cracking, deviations in the stress−strain curve were taken as the primary indicator of the failure strain. The toughness was then calculated by integrating the area under the stress−strain curve up to the failure strain. The elastic modulus of the sample was determined by performing a linear regression on the stress− strain curve between 0.5% and 3% strain, and the strain-to-break was determined as the strain where the sample began to fracture, determined as described in the previous paragraph. Relaxation Testing. Two different testing methods were used for force relaxation experiments (the same sample was never reused for more than one test). The first type of experiment was run up to a lower overall strain (5%) in order to ensure we were probing the linear viscoelastic properties of the sample. The sample was compressed at a constant strain rate of 0.5% s−1 for 10 s, followed by a 5 min holding period. These testing parameters were chosen to be similar to previous protocols that have been described in the literature.13 The second type of experiment was run at a higher overall strain, and was compressed much more quickly in order to approximate a step change in the strain. In this type of test, the sample was compressed at a constant strain rate of 5% s−1 for 5 s, followed by the same 5 min holding period. These latter samples reached a strain of 25%, which is outside the linear viscoelastic region. Nevertheless, comparable strains can be reached in tissue phantoms developed for palpation models,15 and in our case these tests allowed us to empirically model the relaxation of samples closer to their fracture point in order to gain some insight into the mechanisms leading to enhanced toughness at high strains. Modeling. All modeling was done in MATLAB by varying a given model’s parameters to minimize the sum of the squared errors between experimental data and the model fit. Because of random noise in the force reading, the force was not always zero when data collection began. Before running the fitting algorithm, all data was shifted up or down such that the first point was equal to zero. The magnitude of this shift was never more than ±0.002 N. Because of the coupled nature of the E and τ parameters in the KVFD and FM models, these parameters are very sensitive to the initial values supplied to the algorithm. The initial value for τ was always set to 150 s (approximately half of the total testing time). The parameters E and α were manually varied to approximate the fit, and these approximations were supplied to the algorithm as the initial values for the parameters. The experimental data was also fit with the GM model, which contains pure Newtonian dashpot elements, rather than the fractional forms present in the previously described models. A GM model with one Maxwell arm (GM-3) is more commonly known as the Standard Linear Solid or Zener model.14 Finally, the data was also fit with a GM model containing two Maxwell arms (GM-5). This latter model was used to observe the relative advantage that could be gained by allowing more fitting parameters (five, as opposed to the three parameters each of the other models contain) into a standard linear viscoelastic model. It should be noted that a different technique was necessary when modeling the data taken from the high strain experiments. At the strain at which these tests were performed the samples no longer displayed a linear stress response in the ramp region, which made fitting the entire ramp-hold stress responses difficult. Instead, a relatively fast ramp was used to approximate a step change in strain, and the holding period was modeled with a five-term Prony series, which represents a GM-5 model response following a step change in strain. To get more accurate fits in

The fractional elements, represented by the diamonds in Figure 2, therefore have two associated parameters: the viscosity η and the derivative order α. Note that for an α value of 0, eq 4 reduces to eq 1, indicating a purely elastic (Hookean) response, whereas for α equal to 1, eq 4 reduces to eq 2, indicating a purely viscous (Newtonian) response.12 Intermediate values of α describe a hybrid state of viscoelastic behavior; the nature of eq 4 thus allows it to capture a wide array of response behaviors. Analyzing a material’s viscoelastic response is typically accomplished by straining the sample and comparing the theoretical model and experimental responses. Theoretical stress responses were computationally generated via a convolution integral of each model’s relaxation modulus, G(t) (which is equivalent to the material’s stress relaxation for a unit step strain deformation), with the strain rate as follows:11 σ (t ) =

∫0

t

G(t − t ′)

dϵ dt ′ dt ′

(5)

In this paper, a ramp-hold function was used for strain, as has previously been recommended,13 where

⎧ ϵ0 ⎪ t , t < tR ϵ(t ) = ⎨ t R ⎪ ⎩ ϵ0 , t ≥ t R

(6)

Functional forms of G(t) have been previously given for each of the models that we examined,11 so that closed form solutions for the predicted stress relaxation could be obtained and fit to the experimental data by adjusting the model parameters in order to minimize the sum of squared errors (SSE), as described further in the following sections.

■

MATERIALS AND METHODS

Materials. The 250 Bloom Type B gelatin powder was obtained from Custom Collagen (Addison, IL). Alginic acid sodium salt was purchased from Sigma-Aldrich. Germall Plus Liquid was purchased from Crafter’s Choice (Cleveland, OH). Ultrapure (Type 1) DI water was obtained from a Direct-Q 3UV water purification system. Sample Preparation. Gelatin powder was mixed with DI water at room temperature and heated in a 70 °C water bath for 1 h, stirred every 5 min. After the hour, makeup water was added to achieve a final gelatin concentration of 5 wt %, as well as the desired amount of sodium alginate (from 0 to 1 wt %) and a small amount of Germall (0.1 wt %) to prevent microbial growth. Twenty grams of the final solution was then poured in a cylindrical PVC mold, the inner walls of which had been sprayed with mold release spray (MR 311 by Sprayon) to facilitate release of the samples. The molds were then covered with parafilm and left to cure at room temperature for 24 h. Samples were removed immediately before testing to minimize any water loss. Final sample dimensions were, on average, 35 mm in diameter and 21 mm tall. Toughness and Stiffness Testing. All testing was done with a Shimadzu EZ-SX Texture Analyzer equipped a 200 N max load cell and a cylindrical testing head of larger diameter than the samples. Samples to be tested were removed from the molds and their exact dimensions were measured. The bottom plate was covered with water, and additional water was placed on top of the sample in order to facilitate free-slip conditions at both plates. The initial point of contact between the sample and the plate was determined as follows. Before the start of testing, the testing head was cleaned of any debris and dried, and the force reading was tared. To ensure good contact with the surface, the sample was then compressed at 0.02 mm s−1 until a force of 0.05 N was read, which corresponded to a maximum strain of approximately 2%. This method also allowed us to ensure the plates were in good contact with both surfaces of the cylinder, which typically had a small ( t R ⎪ t τn ⎠ ⎝ τn ⎠⎠ ⎝ ⎩ n=1 R ⎝

(11)

For the KVFD model: 3179

DOI: 10.1021/acsbiomaterials.7b00321 ACS Biomater. Sci. Eng. 2017, 3, 3176−3182

Article

ACS Biomaterials Science & Engineering

Figure 3. Model fits of a single data set using the (a) KVFD, (b) FM, (c) GM-3, and (d) GM-5 models. Inset images are zoomed in on the end of the ramp portion of the data. Data in plots are downsampled by a factor of 10 in the ramp region and 100 in the hold region for visualization purposes. ⎧ ⎛ t −α ⎞ ⎪ E ϵ 0t ⎜ ⎟ τ ⎪ ⎜⎜1 + Γ(2 − α) ⎟⎟, t ≤ t R t R ⎪ ⎠ ⎝ ⎪ σKVFD(t ) = ⎨ t − t −α ⎛ t −α ⎞ ⎪ (t − t R ) τ R ⎟ ⎪ E ϵ0 ⎜ τ +t , t > tR ⎪ t ⎜⎜t R − Γ(2 − α) Γ(2 − α) ⎟⎟ ⎪ R ⎝ ⎠ ⎩

element models have a much greater ability to model the ramphold stress relaxation in these systems. The ability of three-parameter fractional calculus models to better fit the stress relaxation data suggests that stress relaxation in gelatin-based hydrogel systems benefit from the flexibility of models with a tunable degree of viscoelastic character, reflected here in the parameter α. Fractional α values in such models can be thought of as generating a stress relaxation response with a continuous distribution of time constants, instead of discrete values.10 Gelatin samples, in particular, comprise a random network of hydrogen-bonded cross-links. The addition of sodium alginate, which may be electrostatically cross-linked through the presence of residual divalent cations in the gelatin, would presumably yield an interpenetrating or biphasic network structure capable of viscous relaxations over a spectrum of time scales, as has similarly been observed in other polymeric systems.7 Given these considerations, it is no surprise that the GM-3 model, which predicts an exponential stress relaxation at a single time constant, τ = η/E, provided a comparably worse fit for the experimental force relaxation data (Figure 3c). Adding additional “arms” to the GM model can provide a better fit for the data through the inclusion of additional parameters (each arm adds another Young’s modulus and viscosity term, and hence another relaxation time parameter). Figure 3d shows the model fit when using a GM-5 model. Despite a comparable SSE between this five-parameter GM-5 model and the two three-parameter fractional models, the inset image reveals that the GM-5 model

()

(

)

()

(12)

For the FM model: ⎧ Eϵ t ⎛ ⎛ t ⎞α ⎞ ⎪ 0 4α ,2⎜− ⎜ ⎟ ⎟, t ≤ t R ⎝ ⎝τ⎠ ⎠ ⎪ tR ⎪ σFM(t ) = ⎨ E ϵ t ⎛ ⎛ t ⎞α ⎞ ⎛ ⎛ t − t R ⎞α ⎞ E ϵ0 ⎟ ⎟, ⎪ 0 4α ,2⎜− ⎜ ⎟ ⎟ − (t − t R )4α ,2⎜− ⎜ ⎝ ⎝τ⎠ ⎠ ⎝ ⎝ τ ⎠ ⎠ tR ⎪ tR ⎪ t>t ⎩ R

(13)

Since the GM-3, KVFD, and FM models all have three fitting parameters, a direct comparison was done on these three models to determine their relative effectiveness at fitting a given set of stress relaxation data. Predicted stress data was generated using eqs 11−13, and Figure 3 shows an example of the model fits for a single data set. All three models do a reasonable job of fitting the long-term relaxation behavior of the samples (Figure 3a−c). The two fractional models, however, do a much better job of fitting the region of the data immediately following the ramp (insets of Figure 3a−c), which results in a much lower overall sum of squared errors. This indicates that the three parameter fractional3180

DOI: 10.1021/acsbiomaterials.7b00321 ACS Biomater. Sci. Eng. 2017, 3, 3176−3182

Article

ACS Biomaterials Science & Engineering

terms to the total modulus, which can be calculated by adding up all the “E” terms. This plot is shown in Figure 4b, and once again, there appears to be no statistical difference across the different alginate concentrations (p = 0.112). In fact, we could find no statistically relevant trends between any of the modeling parameters across the different alginate concentrations we tested. These results suggest that for the low strains (up to ∼5%) and strain rates (∼0.5% s−1) utilized in our initial ramp-hold tests, that the addition of sodium alginate, while enhancing the toughness of the samples, does not appreciably affect the viscoelastic mechanical properties of the gelatin, which could be of interest for designing test phantoms for particular applications. To further explore the observed toughness enhancement observed in the gelatin samples upon addition of sodium alginate, additional force relaxation experiments were run at a much higher ultimate strain (25%). As noted in the methods section, at larger strain the gelatin samples deviate from linear elastic behavior, precluding modeling the entire ramp-hold strain profile. Instead, a faster strain rate was utilized in the ramp region to provide an abrupt increase in strain to εmax = 25%, and the relaxation in these samples was modeled with a five-term Prony series, which is the analytical solution to the GM-5 model for a step change in strain.

still underestimates the amount of relaxation happening on small time scales (a trend that was usually true across multiple samples), again suggesting the superiority of the fractional models for capturing dynamic stress relaxation in the gelatin material. Despite observed success of the fractional models in modeling stress relaxation in our samples, we found that the computational fitting routines were not straightforward to implement. Without careful selection of starting parameters, the fitting algorithms were capable of converging to yield widely different parameter sets with similar SSE values. Care must be taken to avoid local minima traps and solutions with unrealistic values (very large or small values of E or τ).13 Some consistency was gained by initially adjusting the fitting parameters manually, then using these values as the initial inputs to the fitting algorithm. In some cases, the function tolerance needed to be relaxed to keep the algorithm from iterating past a reasonable solution. In an effort to determine whether the increase in the toughness of gelatin/alginate samples was reflected in a corresponding change in the viscoelastic model parameters, we examined the fit values of α (for the fractional models) as a function of alginate concentration, as this parameter is tied to the degree of viscous relaxation occurring within these samples. As seen in Figure 4a,

E (t ) =

⎛ t ⎞ ⎛ t⎞ σ (t ) = E0 + E1 exp⎜ − ⎟ + E2 exp⎜ − ⎟ ϵmax ⎝ τ2 ⎠ ⎝ τ1 ⎠ (14)

By utilizing the GM-5 model to fit the experimental stress relaxation (or equivalently, modulus relaxation) data for these high-strain samples, we are able to observe the relative contribution of shorter-term and longer-term relaxation effects by restricting the fitting parameters to include only two time constants: τ1 (on the order of 3−5 s), which represents shortterm relaxation effects, and τ2 (on the order of 90−150 s), which represents long-term relaxation effects. Figure 5 shows the results of plotting the metric E1/E2, which represents the ratio of the magnitudes of the short-term and long-term modulus relaxation parameters, versus alginate concentration. As can be seen in Figure 5, E1/E2 increases as the concentration of alginate in the sample increases. Also, the value of E1/E2 for samples without alginate was smaller than for

Figure 4. (a) Average value of α using the KVFD model and (b) average ratio of relaxation moduli to the total modulus using the GM-5 model. Uncertainty bars represent the standard error in the sample mean. For these results, 5−8 samples were tested at each alginate concentration. These tests were performed with a strain rate ramp of 0.5% s−1.

we found no statistical difference between the values of the α parameter for the different alginate concentrations (p = 0.174). The value of α does spike at low alginate concentrations, but we believe that spike to be due to outliers in the data. A similar metric can be calculated for the GM-5 model fits by calculating the ratio of the sum of the coefficients in front of the exponential

Figure 5. Average value of E1/E2 for high strain force relaxation experiments. Uncertainty bars represent the standard error in the sample mean. For these results, 6−8 samples were tested at each alginate concentration. 3181

DOI: 10.1021/acsbiomaterials.7b00321 ACS Biomater. Sci. Eng. 2017, 3, 3176−3182

Article

ACS Biomaterials Science & Engineering ORCID

all alginate-containing samples. This indicates that for samples under large strains, adding alginate to gelatin results in a shift in stress relaxation to shorter time scales. This trend is consistent with the observed toughness enhancement when adding alginate to gelatin samples. Mechanistically, if a sample has the ability to relax more quickly, it can more quickly dissipate energy under strain, which would result in an increase in the sample’s toughness. Furthermore, the observed trend is very similar to the trend observed in the increasing strain-to-break of alginatecontaining samples (Figure 2c), which further supports the hypothesis that such samples have an enhanced fluidic character. Others have shown previously that, for pure gelatin systems, short-term relaxation can be associated with fast moving fluid flows within the hydrogel matrix, and longer-term relaxation can be associated with the restructuring of the matrix itself.1 An increase in the short term relaxation modulus therefore suggests that sodium alginate increases the viscous character of the hydrogel at high strains. In regards to the observed toughness increase, the enhanced viscous character of the sample would presumably also be responsible for energy dissipation at higher strains, which would offset chain relaxation and broken crosslinks, and lead to a higher strain-to-break.

Michael A. Samp: 0000-0003-4917-3458 Adam J. Nolte: 0000-0002-2759-2056 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge Dr. Michael Insana and Yue Wang of UIUC, and Matthew Billingsley, Matthew Conrad, Emily Cottingham, Caitlin Douglas, and Haley O’Neil of RHIT for their assistance. Funding for this research was provided by the NSF under Award No. ECCS-1306808.

■

■

CONCLUSIONS We have demonstrated that adding sodium alginate to aqueousbased gelatin samples can increase their toughness without changing their Young’s moduli, making them useful candidates for tissue-mimicking phantoms where specific mechanical properties are desired. The linear viscoelastic properties of gelatin and gelatin/alginate samples were modeled using both the Generalized Maxwell model and two models based on fractional calculus. In all cases, the fractional models provided better three-parameter fits for stress relaxation in the linear elastic region. The models suggest that at low strains, there is no discernible difference in the viscoelastic properties between gelatin and gelatin/alginate samples. At high strains, the observed toughness increase is correlated with a relative increase in shortterm relaxation in the samples, suggesting that alginate modifies short-term fluid flow within the gelatin matrix, allowing for more effective stress dissipation. Although previous studies have indicated that the addition of sodium alginate to gelatin can increase the toughness of dried films,9 our study is the first to demonstrate that small additions of alginate to gelatin hydrogels can provide a significant enhancement in toughness for lowmodulus gelatin hydrogels, while leaving the Young’s modulus unchanged. Our results could be useful in improving the durability of low-modulus tissue phantom materials for biological and medical diagnostic techniques.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsbiomaterials.7b00321. Stress relaxation modeling with ramps having different strain rates, and analysis of ramp speed on the measurement of Young’s modulus (PDF)

■

REFERENCES

(1) Sridhar, M.; Liu, J.; Insana, M. F. Elasticity imaging of polymeric media. J. Biomech. Eng. 2007, 129 (2), 259−272. (2) de Carvalho, I. M.; De Matheo, L. L.; Costa, J. F. S.; Borba, C. D.; von Kruger, M. A.; Infantosi, A. F. C.; Pereira, W. C. D. Polyvinyl chloride plastisol breast phantoms for ultrasound imaging. Ultrasonics 2016, 70, 98−106. (3) Madsen, E.; Zagzebski, J.; Banjavie, R.; Jutila, R. Tissue mimicking materials for ultrasound phantoms. Med. Phys. 1978, 5 (5), 391−394. (4) Sisney, G. A.; Hunt, K. A. A low-cost gelatin phantom for learning sonographically guided interventional breast radiology techniques. AJR, Am. J. Roentgenol. 1998, 171 (1), 65−66. (5) Young, S.; Wong, M.; Tabata, Y.; Mikos, A. G. Gelatin as a delivery vehicle for the controlled release of bioactive molecules. J. Controlled Release 2005, 109 (1−3), 256−274. (6) Czerner, M.; Fasce, L. A.; Martucci, J. F.; Ruseckaite, R.; Frontini, P. M. Deformation and fracture behavior of physical gelatin gel systems. Food Hydrocolloids 2016, 60, 299−307. (7) Sun, J. Y.; Zhao, X. H.; Illeperuma, W. R. K.; Chaudhuri, O.; Oh, K. H.; Mooney, D. J.; Vlassak, J. J.; Suo, Z. G. Highly stretchable and tough hydrogels. Nature 2012, 489 (7414), 133−136. (8) Kong, H. J.; Wong, E.; Mooney, D. J. Independent control of rigidity and toughness of polymeric hydrogels. Macromolecules 2003, 36 (12), 4582−4588. (9) Xiao, C. B.; Liu, H. J.; Lu, Y. S.; Zhang, L. Blend films from sodium alginate and gelatin solutions. J. Macromol. Sci., Part A: Pure Appl.Chem. 2001, 38 (3), 317−328. (10) Lakes, R. S. Viscoelastic Solids; CRC Press: Boca Raton, FL, 1999; p 476. (11) Magin, R. Fractional Calculus in Bioengineering; Begell House, 2006. (12) Zopf, C.; Hoque, S. E.; Kaliske, M. Comparison of approaches to model viscoelasticity based on fractional time derivatives. Comput. Mater. Sci. 2015, 98, 287−296. (13) Zhang, H. M.; Wang, Y.; Insana, M. F. Ramp-hold relaxation solutions for the KVFD model applied to soft viscoelastic media. Meas. Sci. Technol. 2016, 27 (2), 11. (14) Mainardi, F.; Spada, G. Creep, relaxation and viscosity properties for basic fractional models in rheology. Eur. Phys. J.: Spec. Top. 2011, 193 (1), 133−160. (15) Li, M.; Konstantinova, J.; Xu, G. H.; He, B.; Aminzadeh, V.; Xie, J.; Wurdemann, H.; Althoefer, K. Evaluation of stiffness feedback for hard nodule identification on a phantom silicone model. PLoS One 2017, 12 (3), e0172703.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. 3182

DOI: 10.1021/acsbiomaterials.7b00321 ACS Biomater. Sci. Eng. 2017, 3, 3176−3182